[Bug fortran/66480] FPE calling cpu_time() with -ffpe-trap=inexact

[Bug fortran/66480] FPE calling cpu_time() with -ffpe-trap=inexact

[nicolasbock at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=66480

--- Comment #1 from Nicolas Bock <nicolasbock at gmail dot com> ---
gcc-4.7.4 fails just the same.

[Bug fortran/66480] New: FPE calling cpu_time() with -ffpe-trap=inexact

[nicolasbock at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=66480

            Bug ID: 66480
           Summary: FPE calling cpu_time() with -ffpe-trap=inexact
           Product: gcc
           Version: 4.8.4
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: fortran
          Assignee: unassigned at gcc dot gnu.org
          Reporter: nicolasbock at gmail dot com
  Target Milestone: ---

Created attachment 35733
  --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=35733&action=edit
The test program

The attached code, compiled with

$ gfortran -ffpe-trap=inexact test.F90 

results in a SIGFPE:

$ ./a.out 

Program received signal SIGFPE: Floating-point exception - erroneous arithmetic
operation.

Backtrace for this error:
#0  0x7FA644B29757
#1  0x7FA644B29D5E
#2  0x7FA644059EEF
#3  0x7FA644BF8D82
#4  0x4008A8 in MAIN__ at test.F90:?
Floating point exception

[Bug fortran/66480] FPE calling cpu_time() with -ffpe-trap=inexact

[nicolasbock at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=66480

--- Comment #2 from Nicolas Bock <nicolasbock at gmail dot com> ---
gcc-4.9.2 fails as well.

[Bug fortran/66480] FPE calling cpu_time() with -ffpe-trap=inexact

[jvdelisle at charter dot net]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=66480

jvdelisle at charter dot net changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |jvdelisle at charter dot net

--- Comment #3 from jvdelisle at charter dot net ---
It occurs here:

Program received signal SIGFPE, Arithmetic exception.
0x00007ffff7bb291a in _gfortrani_cpu_time_4 (time=0x7fffffffde7c)
    at ../../../trunk/libgfortran/intrinsics/cpu_time.c:53
53        *time = sec + usec * GFC_REAL_4_LITERAL(1.e-6);

[Bug fortran/66480] FPE calling cpu_time() with -ffpe-trap=inexact

[kargl at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=66480

kargl at gcc dot gnu.org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |RESOLVED
                 CC|                            |kargl at gcc dot gnu.org
         Resolution|---                         |WONTFIX

--- Comment #4 from kargl at gcc dot gnu.org ---
It is almost impossible to do any floating point
computation where an FE_INEXACT exception will 
not be raised.  The workaround is don't use
the option that you used.

[Bug fortran/66480] FPE calling cpu_time() with -ffpe-trap=inexact

[nicolasbock at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=66480

--- Comment #5 from Nicolas Bock <nicolasbock at gmail dot com> ---
Thanks for the comments. I should have understood this exception better before
posting this as a bug.

[Bug fortran/66480] FPE calling cpu_time() with -ffpe-trap=inexact

[sgk at troutmask dot apl.washington.edu]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=66480

--- Comment #6 from Steve Kargl <sgk at troutmask dot apl.washington.edu> ---
On Wed, Jun 10, 2015 at 05:20:58PM +0000, nicolasbock at gmail dot com wrote:
> https://gcc.gnu.org/bugzilla/show_bug.cgi?id=66480
> 
> --- Comment #5 from Nicolas Bock <nicolasbock at gmail dot com> ---
> Thanks for the comments. I should have understood this exception
> better before posting this as a bug.
> 

We would rather see a bug report and then simply close it
than to get no bug reports at all.  It is unlikely that
one can avoid an inexact FPE in code involving floating
point unless the algorithm is specifically designed to
avoid it.

[Bug fortran/66480] FPE calling cpu_time() with -ffpe-trap=inexact

[nicolasbock at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=66480

--- Comment #7 from Nicolas Bock <nicolasbock at gmail dot com> ---
(In reply to Steve Kargl from comment #6)
> On Wed, Jun 10, 2015 at 05:20:58PM +0000, nicolasbock at gmail dot com wrote:
> > https://gcc.gnu.org/bugzilla/show_bug.cgi?id=66480
> > 
> > --- Comment #5 from Nicolas Bock <nicolasbock at gmail dot com> ---
> > Thanks for the comments. I should have understood this exception
> > better before posting this as a bug.
> > 
> 
> We would rather see a bug report and then simply close it
> than to get no bug reports at all.  It is unlikely that
> one can avoid an inexact FPE in code involving floating
> point unless the algorithm is specifically designed to
> avoid it.

Thanks :) I agree with you, from it's definition, the inexact result exception
is going to be raised for most operations. Why would one want to trap it?